Scholarlink

This work presents a new model that extends the hubs weighted graphical lasso to dynamic settings by combining it with a Hidden Markov framework. The method is designed to track changes in the network structure over time, especially when hub nodes are present. A penalized EM algorithm is used for estimation, and simulations suggest notable improvements over other HMM-based approaches. Future work will explore how the model behaves when the number of states is overestimated.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating Hub Structures in Hidden Markov Graphical Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Beatrice Foroni,
  • Abbas Khalili,
  • Lea Petrella,
  • Nicola Salvati

摘要

This work presents a new model that extends the hubs weighted graphical lasso to dynamic settings by combining it with a Hidden Markov framework. The method is designed to track changes in the network structure over time, especially when hub nodes are present. A penalized EM algorithm is used for estimation, and simulations suggest notable improvements over other HMM-based approaches. Future work will explore how the model behaves when the number of states is overestimated.